Real-time anomaly detection and parameter adaptive method for precision motion control loop

Abstract

The present application relates to a kind of real-time exception detection and parameter self-adapting method of precision motion control loop, the predicted output value of motion equipment is generated by pre-constructed Kalman filter model, and actual measurement value is obtained, and real-time difference is calculated;When the real-time difference exceeds preset exception threshold value, the process noise covariance parameter of Kalman filter model is dynamically adjusted;According to the process noise covariance parameter after adjustment and the real-time difference, the proportional coefficient and integral time parameter of motion equipment control loop are adaptively adjusted, to realize real-time exception detection and parameter self-adapting adjustment in precision motion control loop, the purpose of improving the control stability and anti-interference ability of high-precision equipment under dynamic disturbance.

Description

Technical Field

[0001] This invention relates to the field of industrial automation and motion control technology, and in particular to a real-time anomaly detection and parameter adaptation method for a precision motion control loop. Background Technology

[0002] In the fields of Industrial Internet of Things (IIoT) and high-precision motion control, physical layer attacks (such as electromagnetic interference and mechanical disturbances) and time-varying characteristics of equipment (such as mechanical wear and temperature drift) pose severe challenges to motion stability. Traditional methods mostly focus on network layer access control or static parameter optimization, lacking the ability to perceive and suppress dynamic anomalies at the physical layer in real time. Existing technologies, such as those using meme search algorithms to optimize layout, cannot cope with control instability caused by real-time disturbances; anomaly detection schemes based on fixed thresholds or rule bases are prone to misjudgment or missed detection due to a lack of dynamic adaptation to noise characteristics and anomaly intensity, making it difficult to meet the requirements of micron-level precision control. In addition, adaptive adjustment of control parameters usually relies on empirical parameter tuning or offline calibration, making it difficult to achieve real-time closed-loop optimization under abnormal operating conditions, which seriously restricts the reliability and intelligence level of high-precision equipment. Summary of the Invention

[0003] The main objective of this invention is to provide a real-time anomaly detection and parameter adaptive method for a precision motion control loop. By dynamically coupling Kalman filter anomaly perception and PID control optimization, the invention aims to achieve real-time anomaly detection and parameter adaptive adjustment in the precision motion control loop, thereby improving the control stability and anti-interference capability of high-precision equipment under dynamic disturbances.

[0004] To achieve the above objectives, the present invention provides a real-time anomaly detection and parameter adaptation method for a precision motion control loop, comprising the following steps:

[0005] The predicted output value of the motion device is generated by a pre-built Kalman filter model, and the actual measured value is obtained to calculate the real-time difference.

[0006] When the real-time difference exceeds the preset abnormal threshold, the process noise covariance parameter of the Kalman filter model is dynamically adjusted.

[0007] Based on the adjusted process noise covariance parameter and the real-time difference, the proportional coefficient and integral time parameter of the motion device control loop are adaptively adjusted.

[0008] Furthermore, the steps for calculating the real-time difference include:

[0009] The prediction channel of the dual-channel Kalman filter is used to predict the operating data of the motion equipment in real time and generate a predicted output value.

[0010] Simultaneously, the actual measurement values ​​of the motion equipment are collected through the actual measurement channel.

[0011] Calculate the real-time difference between the predicted output value and the actual measured value.

[0012] Furthermore, the steps for constructing the Kalman filter model include:

[0013] Based on the historical operating data of the aforementioned motion equipment, the state transition matrix and the observation matrix are fitted using the least squares method.

[0014] Initialize the process noise covariance matrix and the measurement noise covariance matrix. The initial value of the process noise covariance is calibrated based on the maximum acceleration of the equipment, and the initial value of the measurement noise covariance is determined based on the sensor accuracy.

[0015] The time-varying parameters of the state transition matrix are dynamically updated using an online learning algorithm.

[0016] Furthermore, the steps for dynamically adjusting the process noise covariance parameter of the Kalman filter model include:

[0017] When the real-time difference exceeds a preset abnormal threshold, an abnormal signal is triggered.

[0018] Based on the abnormal signal identifier and the squared value of the real-time difference, the process noise covariance parameter of the Kalman filter model is dynamically adjusted, and the adjustment range is positively correlated with the squared value of the real-time difference.

[0019] Furthermore, the step of adaptively adjusting the proportional coefficient and integral time parameters of the motion device control loop includes:

[0020] Based on the adjusted process noise covariance parameter and the real-time difference, adjust the proportional coefficient and integral time parameter of the motion device control loop;

[0021] The adjustment range of the proportional coefficient is positively correlated with the integral value of the real-time difference, and the adjustment range of the integral time parameter is negatively correlated with the absolute value of the real-time difference.

[0022] Furthermore, after the step of adaptively adjusting the proportional coefficient and integral time parameters of the motion device control loop, the following steps are also included:

[0023] Boundary constraint verification is performed on the adjusted proportional coefficient and integral time parameters;

[0024] Optimized control parameters are generated and loaded into the motion device execution unit.

[0025] Furthermore, following the boundary constraint verification step, the following steps are included:

[0026] The scaling factor is limited to a preset range of its initial value; if the scaling factor exceeds the preset range, boundary values ​​are extracted.

[0027] The integration time parameter is limited to a preset minimum time threshold. If the integration time parameter is lower than the minimum time threshold, the preset minimum time threshold is extracted.

[0028] Furthermore, after the step of generating optimized control parameters and loading them into the motion device execution unit, the method further includes:

[0029] The positioning error is calculated based on the difference between the actual position feedback data of the motion device and the target position command;

[0030] When the positioning error exceeds the preset stable range, the dynamic adjustment of noise covariance parameters and the adaptive adjustment of control parameters are re-triggered to form a closed-loop real-time control cycle.

[0031] Furthermore, the steps for calculating the positioning error also include:

[0032] The difference between the actual position feedback data and the target position command is processed by sliding window filtering, and the positioning error is calculated.

[0033] This invention also provides a real-time anomaly detection and parameter adaptation system for a precision motion control loop, comprising:

[0034] The difference calculation unit is used to generate the predicted output value of the motion device through a pre-built Kalman filter model, obtain the actual measurement value, and calculate the real-time difference.

[0035] The parameter adjustment unit is used to dynamically adjust the process noise covariance parameter of the Kalman filter when the real-time difference exceeds a preset abnormal threshold.

[0036] An optimized execution unit is used to adaptively adjust the proportional coefficient and integral time parameter of the motion device control loop based on the adjusted process noise covariance parameter and the real-time difference.

[0037] The real-time anomaly detection and parameter self-adaptation method for precision motion control loops provided by this invention has the following beneficial effects: This invention integrates Kalman filter dynamic detection and real-time self-optimization of PID parameters to construct an integrated control architecture of anomaly perception, parameter adjustment, and closed-loop iteration, overcoming the limitations of traditional methods. Based on a Kalman filter model with dynamic adjustment of noise covariance, it achieves sensitive capture and intensity quantification of weak anomaly signals at the physical layer, significantly improving detection accuracy and anti-interference capability. Through a dual-modal feedback mechanism of anomaly intensity and historical deviation, it drives the nonlinear adaptive adjustment of proportional and integral parameters, balancing transient response speed and steady-state control accuracy. Furthermore, by combining boundary constraint verification and sliding window filtering, it ensures parameter safety and robustness of error criteria, avoiding over-adjustment or false triggering. In addition, through a closed-loop re-triggering mechanism, it forms a continuously self-optimizing control loop, effectively coping with time-varying factors such as equipment aging and environmental disturbances. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating a real-time anomaly detection and parameter adaptation method for a precision motion control loop in one embodiment of the present invention.

[0039] Figure 2 This is a structural block diagram of a real-time anomaly detection and parameter adaptation system for a precision motion control loop in one embodiment of the present invention.

[0040] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0042] Reference Figure 1 This is a flowchart illustrating a real-time anomaly detection and parameter adaptation method for a precision motion control loop proposed in this invention, comprising the following steps:

[0043] S1 generates the predicted output value of the motion device through a pre-built Kalman filter model, obtains the actual measured value, and calculates the real-time difference;

[0044] S2, When the real-time difference exceeds the preset abnormal threshold, the process noise covariance parameter of the Kalman filter model is dynamically adjusted;

[0045] S3. Based on the adjusted process noise covariance parameter and the real-time difference, the proportional coefficient and integral time parameter of the motion device control loop are adaptively adjusted.

[0046] In one embodiment, for step S1,

[0047] The steps for calculating the real-time difference include:

[0048] The prediction channel of the dual-channel Kalman filter is used to predict the operating data of the motion equipment in real time and generate a predicted output value.

[0049] Simultaneously, the actual measurement values ​​of the motion equipment are collected through the actual measurement channel.

[0050] Calculate the real-time difference between the predicted output value and the actual measured value.

[0051] In practical implementation, a Kalman filter framework is constructed based on a state-space model, and its state equation and observation equation are defined as follows:

[0052]

[0053] In the formula, Represents the state vector (position, velocity, acceleration); Indicates the observed value (the actual position measured by the grating ruler); To control the input vector (such as motor drive voltage). These represent process noise and measurement noise, respectively. This represents the pre-trained matrix. The dual-channel Kalman filter generates real-time differences through a parallel processing flow: the prediction channel is based on the state estimate from the previous time step. and control input Calculate the current predicted value and output the predicted location value. The actual measurement channel synchronously acquires the actual position values ​​of the high-precision grating ruler sensor. By comparing the data from the two channels, the deviation between the predicted and actual values ​​is calculated. The threshold was determined through repeated positioning experiments using a semiconductor bonding machine. Covering 95% of abnormal scenarios: as shown in Table 1:

[0054] Table 1:

[0055] Interference type <![CDATA[e k Mean (μm) Standard deviation (μm) No interference 0.8 0.3 Air flotation platform vibration 2.1 1.2

[0056] In one embodiment, the steps for constructing a Kalman filter model include:

[0057] Based on the historical operating data of the aforementioned motion equipment, the state transition matrix and the observation matrix are fitted using the least squares method.

[0058] Initialize the process noise covariance matrix and the measurement noise covariance matrix. The initial value of the process noise covariance is calibrated based on the maximum acceleration of the equipment, and the initial value of the measurement noise covariance is determined based on the sensor accuracy.

[0059] The time-varying parameters of the state transition matrix are dynamically updated using an online learning algorithm.

[0060] Specifically, based on historical operating data, the state-space model parameters are trained offline using the least squares method. The state vector of the motion equipment under normal operating conditions is recorded. (Including position, velocity, acceleration) and control input Sampling period The coefficient matrix of the state equation is solved using the least squares method. and observation matrix :

[0061]

[0062] Taking the X-axis of a semiconductor bonding machine as an example, the fitted state transition matrix reflects the physical relationship between position, velocity, and acceleration; the control matrix correlates motor input and motion response; and the observation matrix directly outputs position measurements, ensuring consistency between the model and actual dynamic characteristics. (Process noise covariance) Based on the equipment's maximum acceleration Calibration (e.g., 0.2 m / s²) is used to extrapolate position and velocity noise components. For example, with a sampling period of 1 ms, the standard deviation of position noise is 0.1 μm, and velocity noise is 0.2 mm / s, reflecting the unmodeled dynamics of the system. Measurement noise covariance is also used. The model's robustness to measurement errors is ensured by directly calculating the accuracy (0.1 μm) of the grating ruler sensor. The state transition matrix is ​​dynamically adjusted using recursive least squares (RLS), and the weights of historical and new data are balanced using a forgetting factor (λ=0.99). Experimental verification is conducted by simulating sudden load changes (from 0g to 50g) on ​​a semiconductor bonding machine, and the prediction errors before and after the model update are compared, as shown in Table 2.

[0063] Table 2:

[0064] condition Mean prediction error (μm) Standard deviation of error (μm) Fixed model (not updated) 3.2 1.8 Online model update 0.9 0.4

[0065] This embodiment addresses the limitations of traditional Kalman filtering, which relies on prior models, by using least-squares offline fitting, thereby improving the consistency between the model and the actual dynamics of the device. The physical calibration of noise covariance avoids overfitting or underfitting caused by empirical assignment, ensuring anomaly detection sensitivity. Furthermore, online updates using RLS are used to address time-varying disturbances (such as mechanical wear and temperature drift) to maintain long-term detection accuracy.

[0066] In one embodiment, for step S2,

[0067] The steps for dynamically adjusting the process noise covariance parameter of the Kalman filter model include:

[0068] When the real-time difference exceeds a preset abnormal threshold, an abnormal signal is triggered.

[0069] Based on the abnormal signal identifier and the squared value of the real-time difference, the process noise covariance parameter of the Kalman filter model is dynamically adjusted, and the adjustment range is positively correlated with the squared value of the real-time difference.

[0070] In practice, when the real-time difference An abnormal signal is generated when the error exceeds a preset abnormal threshold (3 μm, calibrated by repeated positioning experiments on a semiconductor bonding machine). This indicates an abnormal state. In this abnormal state, the process noise covariance is non-linearly amplified based on the squared difference value. The noise covariance is dynamically adjusted according to the following rules:

[0071]

[0072] In the formula, The adjusted process noise covariance reflects the dynamic uncertainty of the system; To measure the noise covariance (a fixed value, preset according to the accuracy of the grating ruler) ); The coefficients are empirical and calibrated through a grid search of historical outlier data to balance sensitivity and stability. The squared term of the real-time difference amplifies the impact weight of strong disturbances. In a simulated physical attack (step disturbance) scenario, the control effects of dynamic adjustment and fixed noise parameters are compared: as shown in Table 3.

[0073] Table 3:

[0074] index Fixed Q parameter <![CDATA[Dynamic adjustment of Q k > Improvement rate Overshoot 12% 2% 83.3% Stabilization time 120 ms 40 ms 66.7% Standard deviation of positional fluctuation 1.8 μm 0.5 μm 72.2%

[0075] This embodiment achieves dynamic matching between anomaly intensity and model sensitivity through a dynamic adjustment rule for noise covariance, suppressing overshoot within a microsecond-level response time, and providing core assurance for real-time disturbance rejection and parameter self-optimization in high-precision motion control.

[0076] In one embodiment, for step S3,

[0077] The steps for adaptively adjusting the proportional coefficient and integral time parameters of the motion equipment control loop include:

[0078] Based on the adjusted process noise covariance parameter and the real-time difference, adjust the proportional coefficient and integral time parameter of the motion device control loop;

[0079] The adjustment range of the proportional coefficient is positively correlated with the integral value of the real-time difference, and the adjustment range of the integral time parameter is negatively correlated with the absolute value of the real-time difference.

[0080] In practice, based on the adjusted process noise covariance (Reflecting the intensity of the anomaly) and real-time difference The control parameters are adjusted using a nonlinear mapping rule, and the proportional coefficient adjustment formula is as follows:

[0081]

[0082] Integral term Accumulated historical deviation, driving Incrementing to accelerate correction, and adjusting the factor. Follow Increase and dynamically improve ( This enhances the strength of the response to persistent anomalies.

[0083] The formula for adjusting the integral time is:

[0084]

[0085] Real-time difference absolute value Suppressing integration time avoids integral saturation caused by large instantaneous deviations, and and A positive correlation ensures that the higher the anomaly intensity, the more significant the decay of the integral effect.

[0086] Experimental verification and performance comparison were conducted by simulating a step disturbance (target position abruptly changed by 10 μm) on the motion axis of the semiconductor bonding machine, and comparing the control effects of traditional PID and adaptive PID, as shown in Table 4.

[0087] Table 4:

[0088] index Traditional PID Adaptive PID (this solution) Improvement rate Overshoot 12% 2% 83.3% Settling time (ms) 120 40 66.7% Steady-state error (μm) ±1.5 ±0.5 66.7%

[0089] Compared to traditional PID control, adaptive PID reduces overshoot from 12% to 2% (an 83.3% improvement), shortens settling time from 120 ms to 40 ms (a 66.7% improvement), and reduces steady-state error from ±1.5 μm to ±0.5 μm. This embodiment overcomes the limitations of fixed parameters in traditional PID control by using a dual-modal control rule of "integral accumulation-instantaneous deviation" combined with anomaly intensity feedback from Kalman filtering. This reduces overshoot and shortens settling time in micrometer-level precision control.

[0090] In one embodiment, after the step of adaptively adjusting the proportional coefficient and integral time parameters of the motion device control loop, the method further includes:

[0091] Boundary constraint verification is performed on the adjusted proportional coefficient and integral time parameters;

[0092] Optimized control parameters are generated and loaded into the motion device execution unit.

[0093] Specifically, in completing the proportional coefficient With integration time After adaptive adjustment, boundary constraint verification is performed. Limit to initial value The range is 0.5 to 2 times (i.e. This prevents over-adjustment from causing oscillations. If the value exceeds the range, the nearest boundary value (such as the calculated value) is used. When forced to be set Set minimum integration time To avoid overshoot caused by excessive integral action. If Then reset to The verified parameters are encapsulated into control commands and transmitted to the motion controller, drive the motor, or actuator via a real-time communication protocol (such as EtherCAT). This implementation uses boundary constraint verification to prevent system instability caused by parameter overshoot, ensuring the physical realizability of the control commands.

[0094] In one embodiment, after the boundary constraint verification step, the following is included:

[0095] The scaling factor is limited to a preset range of its initial value; if the scaling factor exceeds the preset range, boundary values ​​are extracted.

[0096] The integration time parameter is limited to a preset minimum time threshold. If the integration time parameter is lower than the minimum time threshold, the preset minimum time threshold is extracted.

[0097] Specifically, this embodiment further defines a scaling factor truncation rule for the boundary constraint verification step, limiting the scaling factor's value range to its initial value. Between 0.5 and 2 times: if the calculated Less than 0.5 Then it will be forced to be set to 0.5. (Lower limit); if Greater than 2 Then force it to be set to 2 If the value falls between these two values, the original value is retained. For the integration time reset rule, a minimum integration time is preset to prevent excessive integration from causing system oscillations: if the calculated integration time... Less than Then force set to ;like If the original value is not obtained, the parameter over-adjustment is avoided through boundary constraints, thus ensuring the stability of the control system.

[0098] In one embodiment, after the step of generating optimized control parameters and loading them into the motion device execution unit, the method further includes:

[0099] The positioning error is calculated based on the difference between the actual position feedback data of the motion device and the target position command;

[0100] When the positioning error exceeds the preset stable range, the dynamic adjustment of noise covariance parameters and the adaptive adjustment of control parameters are re-triggered to form a closed-loop real-time control cycle.

[0101] Specifically, after the control parameters are loaded into the execution unit, the actual position is based on feedback from the grating ruler. With target position command Calculate the real-time positioning error: .when When the preset stable range is reached, the recalculation of the prediction-measured difference in step S1, the dynamic adjustment of the noise covariance parameter in step S2, and the updating of the control parameters and loading of the process in step S3 are triggered again to achieve closed-loop adaptation.

[0102] In one embodiment, the step of calculating the positioning error further includes:

[0103] The difference between the actual position feedback data and the target position command is processed by sliding window filtering, and the positioning error is calculated.

[0104] Specifically, a sliding window filter is added to the positioning error calculation, using the window width... The moving average filter is applied, and the filtered error is calculated:

[0105]

[0106] In the formula, This refers to the current moment (time step). For the time index within the sliding window, from arrive Covering the most recent At that moment; For the first The positioning error at each moment is used to determine whether to re-trigger the adjustment process based on the filtered error, avoiding interference from factors such as sensor noise that could cause false triggering due to instantaneous noise.

[0107] Reference Figure 2 Here is a structural block diagram of a real-time anomaly detection and parameter adaptation system for a precision motion control loop according to an embodiment of the present invention, comprising:

[0108] The difference calculation unit is used to generate the predicted output value of the motion device through a pre-built Kalman filter model, obtain the actual measurement value, and calculate the real-time difference.

[0109] The parameter adjustment unit is used to dynamically adjust the process noise covariance parameter of the Kalman filter when the real-time difference exceeds a preset abnormal threshold.

[0110] An optimized execution unit is used to adaptively adjust the proportional coefficient and integral time parameter of the motion device control loop based on the adjusted process noise covariance parameter and the real-time difference.

[0111] For the specific implementation of each unit in the above device example, please refer to the method embodiments described above, and will not be repeated here.

[0112] In summary, this invention generates predicted output values ​​of a moving device using a pre-constructed Kalman filter model, obtains actual measured values, and calculates real-time differences. When the real-time difference exceeds a preset anomaly threshold, the process noise covariance parameter of the Kalman filter model is dynamically adjusted. Based on the adjusted process noise covariance parameter and the real-time difference, the proportional coefficient and integral time parameters of the moving device control loop are adaptively adjusted to achieve real-time anomaly detection and adaptive parameter adjustment in the precision motion control loop, thereby improving the control stability and anti-interference capability of high-precision equipment under dynamic disturbances.

[0113] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the present invention and embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0114] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0115] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A real-time anomaly detection and parameter adaptation method for a precision motion control loop, characterized in that, Includes the following steps: The predicted output value of the motion device is generated by a pre-built Kalman filter model, and the actual measured value is obtained to calculate the real-time difference. When the real-time difference exceeds a preset abnormal threshold, the process noise covariance parameter of the Kalman filter model is dynamically adjusted, including: when the real-time difference exceeds the preset abnormal threshold, an abnormal signal is triggered, and the process noise covariance parameter of the Kalman filter model is dynamically adjusted based on the abnormal signal and the square of the real-time difference, wherein the adjustment magnitude is positively correlated with the square of the real-time difference. Adaptive adjustment of the proportional coefficient and integral time parameter of the motion device control loop based on the adjusted process noise covariance parameter and the real-time difference includes: adjusting the proportional coefficient and integral time parameter of the motion device control loop based on the adjusted process noise covariance parameter and the real-time difference, wherein the adjustment range of the proportional coefficient is positively correlated with the integral value of the real-time difference, and the adjustment range of the integral time parameter is negatively correlated with the absolute value of the real-time difference.

2. The real-time anomaly detection and parameter adaptation method for a precision motion control loop according to claim 1, characterized in that, The step of calculating the real-time difference includes: The prediction channel of the dual-channel Kalman filter is used to predict the operating data of the motion equipment in real time and generate a predicted output value. Simultaneously, the actual measurement values ​​of the motion equipment are collected through the actual measurement channel. Calculate the real-time difference between the predicted output value and the actual measured value.

3. The real-time anomaly detection and parameter adaptation method for a precision motion control loop according to claim 1, characterized in that, The construction steps of the Kalman filter model include: Based on the historical operating data of the aforementioned motion equipment, the state transition matrix and the observation matrix are fitted using the least squares method. Initialize the process noise covariance matrix and the measurement noise covariance matrix. The initial value of the process noise covariance is calibrated based on the maximum acceleration of the equipment, and the initial value of the measurement noise covariance is determined based on the sensor accuracy. The time-varying parameters of the state transition matrix are dynamically updated using an online learning algorithm.

4. The real-time anomaly detection and parameter adaptation method for a precision motion control loop according to claim 1, characterized in that, After the step of adaptively adjusting the proportional coefficient and integral time parameters of the motion device control loop, the method further includes: Boundary constraint verification is performed on the adjusted proportional coefficient and integral time parameters; Optimized control parameters are generated and loaded into the motion device execution unit.

5. The real-time anomaly detection and parameter adaptation method for a precision motion control loop according to claim 4, characterized in that, Following the boundary constraint verification step, the following steps are included: The scaling factor is limited to a preset range of its initial value; if the scaling factor exceeds the preset range, boundary values ​​are extracted. The integration time parameter is limited to a preset minimum time threshold. If the integration time parameter is lower than the minimum time threshold, the preset minimum time threshold is extracted.

6. The real-time anomaly detection and parameter adaptation method for a precision motion control loop according to claim 4, characterized in that, After the step of generating optimized control parameters and loading them into the motion device execution unit, the method further includes: The positioning error is calculated based on the difference between the actual position feedback data of the motion device and the target position command; When the positioning error exceeds the preset stable range, the dynamic adjustment of noise covariance parameters and the adaptive adjustment of control parameters are re-triggered to form a closed-loop real-time control cycle.

7. The real-time anomaly detection and parameter adaptation method for a precision motion control loop according to claim 6, characterized in that, The step of calculating the positioning error further includes: The difference between the actual position feedback data and the target position command is processed by sliding window filtering, and the positioning error is calculated.

8. A real-time anomaly detection and parameter adaptive system for a precision motion control loop, characterized in that, include: The difference calculation unit is used to generate the predicted output value of the motion device through a pre-built Kalman filter model, obtain the actual measurement value, and calculate the real-time difference. The parameter adjustment unit is used to dynamically adjust the process noise covariance parameter of the Kalman filter when the real-time difference exceeds a preset abnormal threshold. The adjustment includes: triggering an abnormal signal identifier when the real-time difference exceeds the preset abnormal threshold; and dynamically adjusting the process noise covariance parameter of the Kalman filter model based on the square value of the abnormal signal identifier and the real-time difference. The adjustment range is positively correlated with the square value of the real-time difference. An optimized execution unit is used to adaptively adjust the proportional coefficient and integral time parameter of the motion device control loop based on the adjusted process noise covariance parameter and the real-time difference. This includes adjusting the proportional coefficient and integral time parameter of the motion device control loop based on the adjusted process noise covariance parameter and the real-time difference, wherein the adjustment range of the proportional coefficient is positively correlated with the integral value of the real-time difference, and the adjustment range of the integral time parameter is negatively correlated with the absolute value of the real-time difference.

Images